Nodes & connections
Learn how nodes communicate, how data flows through your workflows, and how to run and manage executions in Spaces.
This guide covers the mechanics of building workflows in Spaces. If you are just getting started, head to Getting started first — it walks you through the canvas, Spotlight, and your first workflow.
In this article
- Types of nodes
- How nodes communicate
- Data types
- Connecting nodes
- What you can connect to what
- The node action bar
- Running your workflow
- Workflow states
- Generation history
- Batch processing with Lists
- Multi-generation
- Credits and cost
- Tips and best practices
Types of nodes
Before connecting anything, it helps to know what tools you have. Each node in Spaces does one specific job — some bring content in, others generate something new, and others transform or combine what you already have.
Upload and Media
These are your starting points. Upload lets you bring in files directly from your computer: images, videos, or audio. Media lets you pick content from your personal history, Freepik stock, or the Community library. Both act as inputs that you connect to other nodes down the line.
Text
The Text node is where you write prompts, scripts, or notes. Click the node or start typing, and a toolbar appears with formatting options — headings, bold, italics, lists, and background colors. You can also use it to leave instructions or context for yourself or collaborators.
Once your text is ready, connect it to a generator node. A single Text node can feed multiple generators at once — for example, the same prompt going to both an Image Generator and a Video Generator.
Image Generator
Generates images based on text prompts, reference images, or a combination of both. You can choose the AI model, set the aspect ratio, and control how many images to create in a single run. This is one of the most common starting points for visual workflows.
Want to try it? Open the template Explore how to create Images in Spaces — it gives you a guided setup to start generating visuals right away.
Video Generator
Creates short-form video content from text or image inputs. You can customize the AI model, duration, aspect ratio, resolution, and number of outputs. It is useful when you want to turn a concept or a still image into motion.
Try the template Let's create your first Video in Spaces to walk through the key steps.
Assistant
An AI-powered helper that adapts to whatever you connect to it. The Assistant can generate prompts from text or visual references, describe images and videos, suggest creative directions, and refine your ideas. Think of it as a collaborator that helps you think — connect it anywhere in your workflow where you need a creative nudge or a better prompt.
Image Upscaler
Improves the resolution and quality of any image. You can choose between Creative mode — which interprets and enhances the image with AI — and Precision mode — which upscales while staying closer to the original. You also control the scale factor and engine type.
Camera Angles
Takes a single image and produces new perspectives — front, side, top, or custom camera positions. It can simulate subtle camera movements and define start and end frames for smooth video transitions. This node is great for product photography, interior design, and adding depth to static visuals.
Audio nodes — Voiceover, Sound Effects, Music Generator
Spaces includes three audio nodes that generate sound from text descriptions. Voiceover converts scripts into natural speech with hundreds of AI voices. Sound Effects creates foley and ambient audio. Music Generator composes original tracks. All three produce audio outputs that you can connect to a Video Generator or combine using the Video Audio Mix node. For a deep dive into models, settings, and prompting tips, see Audio Nodes.
Video Audio Mix
Combines multiple audio tracks with a video. This is typically the final step in an audiovisual workflow — connect your voiceover, music, and sound effects along with your video, and the node mixes everything together.
Utility nodes
These help you organize your canvas rather than generate content. List lets you batch-process multiple items through the same workflow. Sticky Notes are great for leaving instructions or context. Stickers add visual markers. Group lets you bundle related nodes together so you can move and organize them as a unit.
How nodes communicate
Every node has ports — the small circles on its edges. Output ports sit on the right side and send data out. Input ports sit on the left side and receive data in. When you draw a line from an output port to an input port, you create a connection — and that connection carries data from one step to the next.
For example, an Image Generator's output image flows through a connection into an Image Upscaler's input, where it gets enhanced automatically. That is the basic idea behind every workflow in Spaces.
A few things to know about ports
- Some input ports are required — the node will not run without them.
- By default, each input port accepts one connection. If you connect a new source, the old one is replaced.
- Some inputs accept multiple connections — for example, a node that takes reference images might accept several at once.
- Output ports have no limit — you can send one output to as many nodes as you need, which is great for creating variations.
Data types
Data flows between nodes in specific types, and each type has its own color. This makes it easy to trace what is moving through your workflow at a glance.
| Type | Color | What it carries |
|---|---|---|
| Image | Purple | Photos, illustrations, AI-generated images |
| Text | Blue | Prompts, scripts, any text content |
| Video | Green | Video clips, AI-generated videos |
| Audio | Orange | Voiceovers, sound effects, music |
| SVG | — | Vector graphics (experimental) |
You can only connect ports of matching types — an image output to an image input, text to text, and so on. If you try to connect incompatible ports, the line will not snap. This keeps your workflows clean and prevents errors before they happen.
Connecting nodes
There are several ways to wire nodes together. Pick whichever feels fastest for you.
Drag from port to port
The most direct way. Click and hold on an output port, drag the line across the canvas, and drop it on a compatible input port. A colored line appears between the two nodes.
Drag to empty space
Not sure which node you need next? Drag from any port into empty canvas space. Spotlight opens automatically, filtered to show only nodes that are compatible with that port. Select one and it is placed and connected in a single step.
Auto-connect
Drop a new node near another node with a compatible port, and Spaces may suggest an automatic connection. This speeds things up when you are building a chain of nodes in sequence.
Good to know
- Connections always go from output to input — never the reverse.
- You cannot connect a node to itself, and Spaces prevents circular loops where data would go around in circles.
- List nodes with multiple items can connect to single-value inputs — Spaces processes each item one by one.
- When data is flowing, connection lines animate to show direction. Batch connections from Lists appear as dashed lines.
Removing a connection
Hover over any connection line and click the scissors icon that appears, or right-click the line and select Delete.
Want to practice? Try the template Mastering Spaces: From Connection to Creation — it is a guided example that walks you through how nodes interact and flow together.
What you can connect to what
Each node type accepts specific inputs and produces specific outputs. Use this table as a quick reference when building your workflows.
| Node | Accepts as input | Produces as output |
|---|---|---|
| Upload / Media | None — source node | Image or Video |
| Text | None — source node | Text |
| Image Generator | Text prompt + optional reference image | Image |
| Video Generator | Text prompt + optional reference image + optional audio | Video |
| Assistant | Text + optional reference image or video | Text |
| Image Upscaler | Image | Image (upscaled) |
| Camera Angles | Image | Image (new perspective) |
| Voiceover | Text | Audio |
| Sound Effects | Text | Audio |
| Music Generator | Text | Audio |
| Video Audio Mix | Video + Audio | Video (with audio) |
Remember that a single output can feed multiple inputs. For example, one Text node can connect to both an Image Generator and a Video Generator to produce different formats from the same prompt.
The node action bar
When you click a node on the canvas, a horizontal action bar appears above it with quick actions. The exact options depend on the node type, but here is what you might see.
| Action | What it does |
|---|---|
| Run | Execute the node and generate output |
| Duplicate | Copy the node on the canvas |
| Delete | Remove the node from the canvas |
| Download | Save the generated image or video to your device |
| Open preview | View the output in full size |
| Edit image | Open the editing tool for visuals |
| Replace creation | Swap the current input with another |
| Reset creation | Clear the output and start fresh |
Running your workflow
Once your nodes are connected, click the Run button on any node. You will see three options depending on what you want to execute.
| Mode | What it does | When to use it |
|---|---|---|
| Run Node | Runs only the selected node, using whatever data is already on its inputs. | You tweaked one setting — like a seed or a model — and want to re-run just that step. |
| Run Workflow | Runs the entire connected chain from start to finish. | You changed something at the beginning, like a prompt, and want fresh results for everything. |
| Run Downstream | Runs the selected node and everything after it, skipping upstream nodes. | You changed a node in the middle and want to re-process from that point without wasting credits on earlier steps. |
How Spaces decides what runs first
You do not need to worry about order — Spaces figures it out automatically. Source nodes like Text and Upload run first since they have no dependencies. Generators run when their inputs are ready. Processors wait for their upstream nodes. And if your workflow branches, independent branches run in parallel whenever possible.
Cancelling a run
Click the Stop button during execution to cancel. Steps that already completed keep their results — only the currently running and pending steps are cancelled.
Workflow states
Each node shows its current state visually, so you can track progress at a glance.
| State | What it means |
|---|---|
| Idle | Ready to run, has not been executed yet |
| Pending | Queued, waiting for upstream nodes to finish |
| Running | Currently processing — AI generation in progress |
| Completed | Done — results are visible on the node card |
| Failed | Something went wrong — click the node to see the error |
| Paused | Waiting for manual review (Evaluator nodes) |
| Cancelled | You stopped the execution manually |
Generation history
Every time you run a node, the result is saved. You never lose previous outputs when you re-run — they are all kept in the node's history.
- Use the arrow buttons on the node card to browse previous generations.
- Each generation preserves the exact settings that were used to create it.
- This makes it easy to compare results or go back to an earlier version you liked better.
Batch processing with Lists
Connect a List node containing multiple items to a generator, and the workflow runs once per item automatically. Results appear as a stack on the output node, and each one is saved to the generation history.
Some ideas for batch processing:
- Batch prompts — write 10 variations of a prompt and generate all of them at once.
- Batch references — feed multiple reference images through the same pipeline.
- A/B testing — compare outputs across different configurations side by side.
Multi-generation
Most generator nodes have a Number of generations control. Set it from 2 to 10 to produce multiple results in a single run. All results appear in the node's history, and you can browse them with the arrow buttons — great for quickly exploring variations without running the node multiple times.
Credits and cost
AI generation nodes consume credits. The credit cost is displayed on the Run button before you execute, so you always know what a run will cost.
What affects the cost:
- Model — different AI models have different per-generation costs.
- Resolution — higher resolution costs more.
- Scale — the upscaling factor affects credit usage.
- Duration — for video and audio generation, longer outputs cost more.
The cost on the Run button updates live as you change settings, so you can experiment before committing. For full details on how credits work across all tools, see AI credits and limits.
Tips and best practices
Discover compatible nodes from a port. Drag from any port into empty space — Spotlight opens pre-filtered to show only nodes that work with that data type. This is the fastest way to build workflows without guessing.
Use Run Downstream when iterating. Changed something in the middle of your chain? Run Downstream saves time and credits by skipping upstream nodes that have not changed.
Branch your workflow for variations. Connect one output to multiple inputs to process the same data in different ways — for example, sending the same prompt to several Image Generators with different models to compare styles.
Use Lists for batch processing. One List of 10 prompts connected to an Image Generator produces 10 images automatically — no manual repetition needed.
Follow the colors. Color-coded connections help you trace data flow at a glance, especially in complex workflows. Dashed lines mean batch data from Lists.
Check errors on failed nodes. The error banner usually tells you exactly what went wrong and how to fix it.
Explore templates to learn. The best way to understand how nodes and connections work together is to open a template and see a real workflow in action. You can edit, customize, and run any template as your own.
Can't find an answer to your question?
Our support team is here to help you with any questions or issues.
Submit a request